Presentation (CTI)

OBJECTIVES: To investigate public preferences for specific service-offering attributes and marginal willingness-to-pay (mWTP) for artificial intelligence (AI)-driven online medical consultation (OMC) service in China.
METHODS: To clarify the attributes and levels of discrete choice experiment (DCE), we conducted literature reviews and in-depth interviews. Following this, a DCE was designed to evaluate public’s preference for AI-driven OMC service in the context of two hypothetical scenarios: chronic disease and acute disease. The attributes include service provider, platform, response time, communication mode, complexity of medical conditions, qualifications of service provider and service fee. A mixed logit model and latent class models were used in the analysis.
RESULTS: In total, 1096 responses were collected in China, with 775 valid responses used for analysis. Service attribute levels with the highest utility were consistent between chronic and acute disease scenarios: healthcare professionals + AI (chronic, 0.242; acute, 0.178), platform provided by tertiary hospitals (chronic, 0.103; acute, 0.070), instant responses (chronic, 0.515; acute, 0.807), text + photos +videos (chronic, 0.052; acute, 0.021), complex medical issues (chronic, 0.062; acute, 0.035), healthcare professionals/ AI with senior levels (chronic, 0.271; acute, 0.256). The attribute with the largest mWTP value was instant responses (RMB 366.93) for chronic disease scenarios, and instant responses (RMB 737.23) for acute disease scenarios. Latent class analysis identified three groups of people who differed in their choice of AI-driven OMC service.
CONCLUSIONS: Quantifying public preference using discrete choice methodology provides information needed to optimize AI-driven OMC service. The study has also demonstrated the importance of response time in the preference of OMC service.